(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Children's Library | Biodiversity Heritage Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
Upload
See other formats

Full text of "The Likert organizational profile: methodological analysis and test of System 4 theory in Brazil"

The Likert Organizational Profile: 
Methodological Analysis and Test of System 4 
Theory in Brazil 



Dewe, 





D. Anthony Butterfield and George F. Farris 
Working Paper No. 608-72 July, 1972 



The Likert Organizational Profile: 
Methodological Analysis and Test of System 4 
Theory in Brazil 

D. Anthony Butterfield and George F. Farris 
Working Paper No. 603-72 July, 1972 




OCT 17 1972 

M. 1. T. LIbKAKltS 



Butterfield 



The Likert Organizational Profile: 
Methodological Analysis and Test of System 4 Theory in Brazil 

Abstract 
A 20-item Likert Organizational Profile (LOP) was administered twice 
to 256 employees in 13 Brazilian development banks. Actual and ideal 
bank profiles were similar to those found in the U.S. and elsewhere: 
employees want participative-group management methods but say their organ- 
ization uses autocratic or consultative methods. Factor analyses did 
not yield the six dimensions predicted by Likert ' s theory; factors were 
only partially consistent over time and for different hierarchical levels. 
Retrospective scores were quite accurate and equally so for 6, 12, and 
18 month time periods since original administration. Test-retest relia- 
bility of the LOP as a whole was moderate. Bank LOP scores were unrelated 
to objective measures of organizational effectiveness, but were positively 
related to employee satisfaction. The LOP appears useful for current 
and retrospective organization studies, but the theory of management systems 
measured by it was only partially supported. 



63G345 



Butterfield 



The Likert Organizational Profile: 
Methodological Analysis and Test of System A Theory in Brazil-' 

D. Anthony Butterfield^ and George F. Farris^ 

Alfred P. Sloan School of Management 
Massachusetts Institute of Technology 

The work of Rensis Likert and his associates at the Institute for 
Social Research represents a major contribution to modern organization 
theory. Likert' s two books (Likert, 1961, 1967) summarize much of the 
work of the Institute and present his own theory of management. The 
theory postulates four systems of management, and argues that system 4, 
"participative-group", is the most effective. A considerable amount of 
empirical research is presented in support of the theory. Most of the 
research was done in organizations in the United States. 

More recently, the theory has been tested in Yugoslavia and Japan, 
using various versions of a "profile of organizational characteristics" 
(Likert, 1969; Kavcic, Rus, & Tannenbaum, 1971). The profile is a ques- 
tionnaire which describes an organization on the dimensions of Likert 's 
theory, and has come to be knovm as the "Likert Organizational Profile", 
or LOP. The LOP has also been used in the United States, frequently in 
studies of change (e.g.. Marrow, Bowers, & Seashore, 1967; Golembiewski 
& Carrigan, 1970a and b; Blumberg & Wiener, 1971). 

Despite this use of the LOP in organization studies, it has not been 
subject to much analysis as a measuring instrument. Little evidence has 



Butterfield 3 

been presented as to the LOP's reliability or factor structure. The 
purpose of the present paper is to present a methodological analysis of 
the LOP, and to examine the suitability of using the LOP to study organ- 
izations in a country outside the United States — Brazil in the present 
case. We shall also test support for Likert's theory in Brazil. 

To meet these objectives we shall a) present profiles of Brazilian 
organizations as seen by their members in terms of Likert's four management 
systems; b) discuss factor analyses of the LOP; c) present a longitudinal 
analysis of the profile; d) examine relationships between LOP scores and 
organizational effectiveness and employee satisfaction. 

Method 
Instrument 

The LOP used in the present study was Form S, distributed by the 
Foundation for Research on Human Behavior (1967). Two items (nos. 2 - sub- 
ordinates confidence in superiors - and 12 - subordinates knowledge of 
organ:, ttional policies) were added at the request of the Brazilians, making 
a tot.il of 20 items on the profile. Questions were translated from English 
into Portuguese by bilingual Brazilians. These translations were then 
checked and revised by the authors in collaboration with the translators 
until all were satisfied that the translations were as accurate as possible. 
An independent re-translation into English indicated that the Portuguese 
and English versions were essentially identical. 

LOP questions are answered on a 20-point scale divided into four 
sections, each section representing one of the four management systems 
described by Likert's theory. Each question has a different set of four 
descriptive alternatives, corresponding to the characteristics of the four 
management systems. For example, question 1 asks, "How much confidence is 



Butterfleld _ , 

shown in subordinates by superiors?" The four alternatives spread out 
on the 20-point scale are, "Very little, Some, Substantial, Very much." 
Responses more toward the right on the scale receive a higher numerical 
value and indicate use of a management system more like the participative- 
group system 4 seen by the theory as most effective. According to Likert 
(1967), the fact tHat all the scales are in the same direction does not 
influence the results. 

The items in Form S are grouped into six theoretical dimensions of 
organizational processes. Items 1-4 measure leadership; 5-7, motivation; 
8 ~.12, communication; 13 - 15, decision making; 16 - 17, goal setting; 
and 18 - 20, control. Both individual item scores (Golembiewski & Carrigan, 
1970a and b; Kavcic, Rus & Tannenbaum, 1971) and index scores (Blumberg & 
Wiener, 1971; Marrow, Bowers, & Seashore, 1967) on the basic dimensions 
have been used in previous research. 

Procedure 

The LOP was administered in group sessions during visits to Brazilian 
development finance institutions. On the first administration, instructions 
were to answer each question twice, once to describe the organization as it 
was actually, and a second time as it should be ideally. On the second 
administration, six to eighteen months later, instructions were to describe 
the organization as it was actually, and as it had been previously (at the 
time of the first administration, although the first administration was not 
referred to specifically). 

Complete data are available from 13 organizations. One was privately 
owned, the others were either development banks or development finance 



Butterfield 

companies, owned by state or regional governments. For two banks there 
was a six-month time lag between administrations; for eight banks there was 
a 12 month time lag; and three banks received the LOP 18 months after the 
original administration. These differences in time were designed to look 
at accuracy of retrospection as a function of the passage of time. It was 
the authors' impression that the participants in this study were quite 
interested in the task of completing the questionnaire; questionnaire 
administration sessions seemed to go as well as they do in the United States 
with similar respondents. 

Respondents 

Respondents were professional employees ranging from non-supervisors 
to bank presidents. Sixty-seven per cent were between the ages of 25 and 
34; 76% had more than 15 years of schooling; 53% had been with their organ- 
izations four years or less; 80% were either lawyers, economists, engineers, 
or accountants. 

Total individual N's for the Time 1 and Time 2 administrations were 
211 and 256, respectively. They represent a non-random sample of about 85% 
of the professional staff involved with a loan program for small and medium 
sized industrial firms. N's within banks varied from 7 to 41, the size 
of the sample generally reflecting the size of the bank. Size did not 
influence the results. 

Analysis 

Except for the factor analyses and hierarchical analyses, data were 
analyzed using bank mean scores. The bank was selected as the unit of 
analysis since the LOP is designed to measure organizational characteristics, 
and our primary concern is with the management systems of organizations. 



Butterfield 

6 

Results 
Profiles 

What management system are these Brazilian development banks actually 
using, and what system would they like to be using ideally? Figure 1 



Insert Figure 1 about here 



presents the average actual and ideal profiles from the 13 banks, based on 
the first LOP administration. It is evident that these 13 financial insti- 
tutions were actually using a system 3 "consultative" style of management, 
with system 2 attributes in the area of decision making and goal setting. 
The ideal style was seen as system 4. Analyses of variance indicated 
essentially no differences among banks on their ideal profiles, and signifi- 
cant differences on their actual profiles. Similarly, analysis of variance 
among three hierarchical levels in the banks (non-supervisory professionals, 
supervisors, and top management) showed no difference on ideal profiles, 
but frequent differences on the actual profiles. Higher organizational levels 
tended to see their organization as more toward system 4 than did lower 
levels. Thus there was substantial agreement among banks and organizational 
levels that system 4, "participative-group", would be the ideal management 
style, yet the actual system was seen as being more towards consultative or 
autocratic. 

The profiles in Figure 1 are similar to those found in Likert's studies. 
They suggest that, in terms of Likert's theory, Brazilian development banks 
may be using management systems not very different from those used by 
organizations in the United States, or even Yugoslavia or Japan. 



Buttcrfield - ^ 

Factor Analysis of the LOP 

The short form of the LOP was designed to measure six basic organi- 
zational processes: leadership, motivation, communication, decision making, 
goal setting, and control. Does it? Factor analyses of the LOP v^ere 
carried out to examine its basic structure and to see if there was empirical 
support for the proposed six dimensions. 

Table 1 presents results of the factor analysis of the ideal responses 
from individual respondents at all organizational levels. A principal 



Insert Table 1 about here 



components analysis was used, with varimax rotation for six factors. 
Except for the first factor, which is clearly leadership, the factors are 
at best mixtures of the dimensions proposed by the theory. We have given 
them the tentative labels of "resistance, guidance, informed decision 
making, dispersion of goal setting and control," and "motivation and 
communication." In additional analyses it was more difficult to label 
factors. Note that analysis should have stopped with five factors, since 
the Eigenvalue for the sixth factor is less than 1.00. A careful examina- 
tion of Table 1 also indicates than neither item 5 (type of motivation used) 
nor item 7 (amount of teamwork) find a place in any of the six factors. 

Additional factor analyses were carried out for the actual responses 
from the two time periods. However, since results were much less inter- 
pretable, they will not be presented here. Controls for organizational 
level tended to complicate rather than clarify results. The most con- 
sistent factors were leadership and resistance, but they emerged in a 
different order and with a different make-up in each analysis. Usually 
there were only three or four factors with Eigenvalues greater than 1.00. 



Butterfield 8 

Items 7 and 5 frequently did not appear. 

The results in Table 1 represent in a sense the best of what the 
factor analyses produced; that is, somewhat interpretable factors with 
at least some relationship to the theory. Repeated analyses produced 
only fair factor consistency over time, fair overlap between ideal and 
actual factors, and fair to poor comparability across organizational levels. 
This lack of consistency in factors across different analyses was the 
characteristic outcome, not the emergence of a large and possibly Halo- 
induced general factor. Other studies (Likert, 1967; Marrow, Bowers & 
Seashore, 1967) have alluded to such a Halo factor, but none was found in 
the present case. Since factors were not clear cut and consistent 
across the different analyses, the creation of factor scores did not seem 
justified. Subsequent analyses were based on individual items as well as 
a total LOP score (overall average across the 20 items) . The latter seemed 
justified in view of high intercorrelations among the twenty LOP items 
and Likert 's (1967) assertion that the LOP measures relatively consistent 
management systems. 

4 
Longitudinal Analysis of the LOP 

How reliable are LOP scores over time? In particular, how accurate 
are LOP scores when used retrospectively? The accuracy issue should be 
of general interest since investigators frequently ask respondents on 
questionnaires to remember how things were at some point in the past, with- 
out knowing in fact how they really had been. In the present study we knew. 

Accuracy of retrospection . We might expect that accuracy of retrospection 
decreases as time increases due to forgetting. Thus in the present study 
the banks in which the Time 2 administration followed the Time 1 adminis- 



Butterfield ^ 

tration by six months should be the most accurate: retrospective scores 
from the 18-month time lag banks should be the least accurate, and the 12 
month group of banks should fall in between. However, if retrospection 
were perfect, the actual and retrospective profiles within each time 
grouping should be the same. Preliminary visual inspection showed re- 
markable overlap between the actual (first administration) and retrospective 
(second administration) profiles y/ithin each of the 3 time groups. T tests 
done on an item by item basis, separately for each time group, confirmed 
that there were few significant differences between actual and retrospective 
item scores, for any time group. These results are summarized, using mean 
total LOP scores, in Table 2. An F test on the average difference score 



Insert Table 2 about here 



for the three groupings confirmed that there was no difference in accuracy 
among the three retrospective periods (F = 0.268). 

Test-retest reliability . To look at "real change" as a function of 
time, similar longitudinal analyses were carried out comparing Time 1 
actual LOP scores with Time 2 actual scores. Results (not shown) were 
the same as for retrospective scores: time had no effect. Since time 
differences produced no differences in either retrospective accuracy or 
change, simple rank order correlations were calculated using all 13 bank 
means on total LOP. Time 1 actual scores correlated rho .63 with Time 2 
retrospective scores, and rho .52 with Time 2 actual scores. With an N of 13', 
both correlations are significant beyond the .05 probability level using 
a one-tail test. The first rho (.63) might be considered the retro- 
spective reliability of the LOP. The second (.52) is the test-retest 



Butterfield 

11 

one another. 

The last four measures in Table 3 are items taken from another 
questionnaire administered in the banks at the time of the second LOP 
administration. They are the employees' ovm perceptions of how effec- 
tive their bank is, how satisfied they are with their own supervisor, 
the extent to which the bank's employees are satisfied rather than dis- 
satisfied, and their evaluation of their bank as a place to work compared 
with other organizations they know. Since organizational level had a 
significant effect on these items (higher levels reporting more favorable 
responses) , the results in Table 3 are based on LOP and questionnaire data 
from non-supervisory professionals only. 

Table 3 shows that bank scores on the LOP were not related to objective 
measures of bank effectiveness, but they were related in the predicted 
manner to three of the four questionnaire measures of effectiveness and 
satisfaction. Contrary to theory, system 4 methods of management were 
not associated with better organizational performance, but they were 
associated with employees' feelings about performance and with employee 
satisfaction. The theory is thus only partially supported. 

Note the nature of the one satisfaction item in Table 3 not related 
to LOP. Apparently, employees who described their organization as more 
towards system 4 were not more satisfied with their own supervisor (nor 
were they more dissatisfied). One's immediate supervisor may not be more 
satisfying in a system 4 organization, yet employees do feel that the entire 
organization is more effective, is a better place to work, and has more 
satisfied employees. It appears that the LOP can distinguish between more 
general organization-wide and more immediate work-group characteristics. 



Butterfield 12 

Discussion 

Development banks are probably among the more modern of Brazilian 
organizations. They are relatively new, and hire a high percentage of 
young and well educated employees. Under these conditions, the LOP appears 
to be a useful instrument for organization studies in Brazil. Questionnaire 
administration went quite smoothly, and profiles were remarkably similar 
to those obtained in the United States. Common stereotypes of Latin 
American culture would p'-edict that in more traditional Brazilian organ- 
izations we might encounter more difficulties in administration of the LOP, 
and we might find both actual and ideal profiles toward the authoritative 
end of the scale. These stereotypes merit empirical test. 

The factor structure of the LOP does not correspond well to the six 
hypothesized organizational processes. This is not a serious drawback, 
but it does suggest that index scores on the sub-scales should be used 
with great caution. Likert himself recognizes the difficulty in obtaining 
factors corresponding exactly to the proposed dimensions (Likert, 1967). 
He argues that either ideal scores or actual scores from within a single 
organization will yield cleaner factors. In the present study we did not 
have sufficient data to justify factor analysis from within a single 
organization. Factor scores aside, much can still be done with individual 
items and especially with a total LOP score. Using ideal and actual 
profiles in a feedback program for organizational development has been 
useful in both Brazil (Butterfield, 1970) and the United States (Blumberg & 
Wiener, 1971). 

The reliability of the LOP over an extended period of time is a complex 
question. On the one hand, it seems quite clear that organizational members 



Butterfield 

13 

do a fairly accurate job in remembering the state of their organization ■ 

,-, 6 
6, 12, or 18 months into the past. Furthermore, these differences in 

time have no effect on the accuracy of retrospection. Using the LOP 

retrospectively thus appears to be quite legitimate, at least up to a 

period of 18 months. On the other hand, total scores on the LOP are 

only moderately stable over time. The test-retest reliability (.52) 

obtained in the present study, though statistically significant, is 

certainly not high by traditional testing standards. However, given the 

small number of organizations (13) and the rather dynamic state of the 

development finance field in Brazil, perhaps we ought not to have expected 

the test-retest reliability to have been much higher. 

Results of the correlations between LOP and dependent variables are 
of particular theoretical interest. As predicted, Brazilian development 
bankers were more satisfied with participative-group management systems 
(although such satisfaction was not dependent on satisfaction with the 
immediate supervisor), and said such systems would be the ideal. Contrary 
to theory, organizations tending to use such systems were not more 
effective in terms of objective performance. 

System 4 theory would explain these different results by invoking 
a time-lag argument. The benefits of using system 4 methods are already 
apparent in the greater satisfaction of organization members. Such 
satisfaction usually precedes changes in organizational end-product variables. 
Improved performance is on the way; it is only a matter of time. Unfor- 
tunately, although we do have longitudinal data in the present study, we 
do not have sufficient data over a sufficient time period to test adequately 
the time-lag hypothesis. 



Butterfield 



14 



An alternative explanation is the "manpower competence-performance 
feedback loop hypothesis" (Farris, 1969; Farris & Butterfield, 1971). 
The effectiveness of participative methods depends on having competent, 
experienced employees. When such a manpower base does not exist, as is 
the likely case in developing countries, participative raetl^ods are not 
likely to be more effective. Greater relative competence resides at the 
supervisory level, so that closer- methods of supervision may be more 
appropriate, as well as "lore accepted (Farris & Butterfield, 1971). In the 
present study, for example, satisfaction with immediate supervisor was 
not significantly related to being in a more participative organization. 

These are alternative rather than competing hypotheses. Results of 
the present study suggest that the LOP may be a useful instrument for 
exploring them further, even if system 4 theory is only partially supported 
in Brazil. 



Butterfield 15 

References 

Blumberg, A., & Wiener, W. One from two: facilitating an organ- 
izational merger. Journal of Applied Behavioral Science, 1971, 
l_y 87-102. 

Butterfield, D. A. Pesquisa de agao em desenvolvimento organ: zacional 
na America Latina pelo metodo de levantamento de feedback. Revista 
de Administragao Publ ica, 1972, 6^, (1), 89-99. 

Farris, G. F. Organizational factors and individual performance: a 

longitudinal study. Journal of Applied Psychology , 1969, _52' 87-92. 

Farris, G. F., & Butterfield, D. A. Are current theories of leadership 
culture-bound? An empirical test in Brazil. Invited paper presented 
at the symposiiim, "Contemporary Developments in the Study of Leader- 
ship", Southern Illinois University at Carbondale, April, 1971. 

Foundation for Research on Human Behavior. Sample kit: Profile of 
organizational characteristics. Ann Arbor, Michigan: Institute 
for Social Research, 1967. 

Golembiewski, R. T. , & Carrigan, S. B. The persistence of laboratory- 
induced changes in organization styles. Administrative Science 
Quarterly , 1970, 15^, 330-3A0. 

Golembiewski, R. T. , & Carrigan, S. B. Planned change in organization 

style based on the laboratory approach. Administrative Science Quarterly , 
1970, 15^, 79-93. 

Kavcic, B. , Rus, V., & Tannenbaum, A. S. Control, participation, and 

effectiveness in four Yugoslav industrial organizations. Administrative 
Science Quarterly, 1971, 16, 74-86. 



Butterf ield 



16 



Likert, R. New patterns of management . New York: McGraw-Hill, 1961. 

Likert, R. The human organization . New York: McGraw-Hill, 1967. 

Likert, R. The relationship between management behavior and social 
structure - improving human performance: Better theory, more 
accurate accounting. Symposium (C-3) presented at the meeting of 
the Conseil International Pour L' Organisation Scientif ique , Tokyo, 
1969. 

Marrow, A. J., Bowers, E G., & Seashore, S. E. Management by partici- 
pation . New York: Harper & Row, 1967. 



Butterfield 

17 

Footnotes 

1 This study was supported by a grant from the Ford Foundation. The 
authors wish to thank members of the Grupo de Trabalho Misto for their 
help, and Dr. Rensis Likert for his permission to use the LOP. 
Special acknowledgement goes to Cecilio Berndsen and Eldon Senner 

for their assistance in gathering and processing the data, 

2 Now at the School of Business Administration, University of Massachusetts, 
Amherst, Massachusetts. 

3 Requests for reprints should be sent to George F. Farris, Sloan School 

of Management, Massachusetts Institute of Technology, Cambridge, Massachu- 
setts, 02139. 

4 All longitudinal analyses were carried out separately for each organi- 
zational level. Results were essentially identical. Data presented 
in the longitudinal section are from respondents at all organizational 
levels combined. 

5 This ratio includes development loans of all types, not just small and 
medium industrial projects. The rank ordering of banks on financial 
data was so highly correlated from year to year in prior years that 
data on this ratio were not obtained for 1970. 



Butterfield 



X8 



TABLE 1 
Factor Analysis of Ideal Likert Organizational Profile 



Factor 



Item 



"Leadership" 

Eigenvalue 5.689 
Cumulative prop, of 
total variance 



1) Confidence in subordinates .799 

2) Confidence in superiors .75A 

3) Subs feel free to talk to sups .714 

4) Subs ideas sought, used .627 



"Resistance" 

Eigenvalue 1.850 
Cumulative prop, of 
total variance .377 



17) Covert resistance org. goals .820 
19) Informal org. resists formal .760 



3. "Guidance" 

Eigenvalue prop, of 
total variance .452 



20) Use of internal control data .737 
9) Acceptance communication down .706 



4. "Informed Decision Making" 

Eigenvalue 1.225 
Cumulative prop, of 
total variance .513 



14) Subs involved decisions 

related to work 

11) Sups know subs' problems 
13) Where decisions made 

15) *D-M contribute motivation 

12) *Subs know org'l policies 



,674 

,662 
,628 
,577 
,500 



"Dispersion of Goal 

Setting and Control" 

Eigenvalue 1.036 
Cumulative prop, of 
total variance .565 



18) Distribution of internal control .663 
16) How org'l goals set .620 



6. "Motivation and Communication" 

Eigenvalue .978 
Cumulative prop, of 
total variance .613 



8) Direction of information flow .760 

6) Where responsibility felt .690 

10) Accuracy of communication up .610 



* Item loads on more than one factor over .40 



Butterfield 



19 



TABLE 2 
Summary: Effect of Time on Retrospection 







Mean 


Total 


LOP 










Time 


Time 1 








Retro- 








Group 


Actual 








spective 


T 


df 


P 


6 months 


9.91 








10.48 


.82 


1 


NS 


12 months 


11.54 








11.46 


-.23 


7 


NS 


18 months 


12.32 








12.49 


.18 


3 


NS 



Butterfield 



20 



TABLE 3 

Correlations Between Bank Total LOP Scores 
and Effectiveness, Satisfaction 



Effectiveness LOP 

1. Number of development loans, 1970 ■ -.01 

2. Ratio of development loans to total assets, 1969 .07 

3. Rankings of overall effectiveness by National Bank 

experts, 1970 .28 

4. Bank self-rating of effectiveness .78** 

Satisfaction 

5. Satisfaction with immediate supervisor .24 

6. Employee satisfaction with bank .77** 

7. Evaluation of bank as place to work .83** 



** p <i..01, one-tail 

Note. - Rank-order correlations using Time 2 actual LOP scores and questionnaire 
data from non-supervisory professionals only. N's range from 11 - 15, 
including 2 banks not part of longitudinal analysis. 



Butterfield ' 21 

Figure Caption 

Figure 1. Bank Mean Ideal and Actual LOP Profiles, Time 1 
(N = 13 banks) 



Butterfield 



22 



1. Confidence in subordinates 

2. Confidence in superiors 

3. Subs, feel free talk sups. 

4. Subs, ideas sought, used 

5. Type of motivation used 

6. Wliere responsibility felt 

7. Amount of teamwork 

8. Direction information flow 

9. Acceptance communication down 

10. Accuracy communication up 

11. Sups, know subs.' problems 

12. Subs, know org.'l policies 

13. Where decisions made 

14. Subs, involved dec. rel. work 

15. D-M contribute motivation 

16. How org.'l goals set 

17. Covert resistance org. goals 

18. Distribution internal control 

19. Informal org. resists formal 

20. Use of internal control data 



System 
2 3 



20 




Actual 



Ideal 



Fig. 1. Bank Mean Ideal and Actual LOP Profiles, Time 1 
(N = 13 banks) 






^^^#. 



.'74^